smalltalk hasn't made it big
Author |
Message |
Zachary William #1 / 8
|
smalltalk hasn't made it big
What has kept Smalltalk from catching on with the broader programming community? Would you consider Smalltalk as the best OO language? I recently read a book by Scott Ambler called 'Building Object Applications That Work' and he convinced me (through words not experience) that Smalltalk was the most pure of all OO programming languages. Are there any drawbacks to the language as a result of the language being designed relatively early to other programs? What say you, present company? Z
|
Wed, 18 Jun 1902 08:00:00 GMT |
|
|
James A. Robertso #2 / 8
|
smalltalk hasn't made it big
There have historically been two problems: 1) When ST first came out (1980 publically), the systems in general use just weren't powerful enough. So Smalltalk got tarred early on for being 'slow' and 'bloated' Interestingly enough, a full VisualWorks (for example) development environment installed (with all optional add ons) takes up less space than a typical MS office install. And the dev environment chews between 6 and 12 MB of RAM (less when you deploy). So the early tarring isn't accurate, but it has stuck to some extent. 2) Very, very poor marketing. ParcPlace never had any idea of what it was doing in the market. Digitalk offended all it's early customers when they raised prices dramatically. Then ParcPlace and DIgitalk merged, and there followed 18 months of confused message and no product releases. Then new management came in and focused on Java. IBM doesn't care what you buy, so long as it's VisualAge (Smalltalk, Java, C++, whatever). However, problem two should start to recede with VisualWorks in good hands (Cincom) and other vendors occupying the space Digitalk left (Object-Arts, ObjectConnect, QKS) Quote:
> What has kept Smalltalk from catching on with the broader programming > community? Would you consider Smalltalk as the best OO language? I > recently read a book by Scott Ambler called 'Building Object Applications > That Work' and he convinced me (through words not experience) that > Smalltalk was the most pure of all OO programming languages. Are there any > drawbacks to the language as a result of the language being designed > relatively early to other programs? > What say you, present company? > Z
-- James A. Robertson Senior Sales Engineer ParcPlace Division of CinCom
Talk Small and Carry a Big Class Library
|
Wed, 18 Jun 1902 08:00:00 GMT |
|
|
Taylor Core #3 / 8
|
smalltalk hasn't made it big
There is something beyond the reasons you cited. One person posted a message here that IBM told a user's meeting that VAST was no longer a primary target for IBM. This prompted Dibbe Edwards to respond saying VAST would be supported in the future. Posts like these indicate to me, if I may be allowed to speculate, that there are forces within IBM that would like the whole VA/Smalltalk product line to go away. I don't know what these people are specifically rejecting. The VA/Smalltalk line currently has over 70K dev licences and this comes from VAST selling itself (IBM does not make VAST sales calls). One can only imagine the kind of sales VAST would have if IBM did active selling. Another interesting point that someone made on a yahoo message board, when IBM developed VA/Java, they used ST as their development tool of choice. IBM, however, recommends VA/Java for their customers who might need to do the same kind of serious development. I understand that IBM is moving VA/Java off Smalltalk but still find it interesting. --tc
Quote: > There have historically been two problems: > 1) When ST first came out (1980 publically), the systems in general use > just weren't powerful enough. So Smalltalk got tarred early on for > being 'slow' and 'bloated' > Interestingly enough, a full VisualWorks (for example) development > environment installed (with all optional add ons) takes up less space > than a typical MS office install. And the dev environment chews between > 6 and 12 MB of RAM (less when you deploy). So the early tarring isn't > accurate, but it has stuck to some extent. > 2) Very, very poor marketing. > ParcPlace never had any idea of what it was doing in the market. > Digitalk offended all it's early customers when they raised prices > dramatically. Then ParcPlace and DIgitalk merged, and there followed 18 > months of confused message and no product releases. Then new management > came in and focused on Java. IBM doesn't care what you buy, so long as > it's VisualAge (Smalltalk, Java, C++, whatever). > However, problem two should start to recede with VisualWorks in good > hands (Cincom) and other vendors occupying the space Digitalk left > (Object-Arts, ObjectConnect, QKS)
> > What has kept Smalltalk from catching on with the broader programming > > community? Would you consider Smalltalk as the best OO language? I > > recently read a book by Scott Ambler called 'Building Object Applications > > That Work' and he convinced me (through words not experience) that > > Smalltalk was the most pure of all OO programming languages. Are there any > > drawbacks to the language as a result of the language being designed > > relatively early to other programs? > > What say you, present company? > > Z > -- > James A. Robertson > Senior Sales Engineer > ParcPlace Division of CinCom
> Talk Small and Carry a Big Class Library
|
Wed, 18 Jun 1902 08:00:00 GMT |
|
|
pharsei #4 / 8
|
smalltalk hasn't made it big
There's one last point that hasn't been covered here. While in most ways, the {*filter*} computer languages have been evolving toward the principles of Smalltalk, they remain strongly typed while Smalltalk uses run-time type checking. I suppose this is primarily a result of CPU's not having primitive type checking ops ( and maybe just inertia ). Specialized CPU's with tagged data structures and primitive type checking have been implemented (The Design and Evaluation of a High Performance Smalltalk System details one), but the idea hasn't really caught on. When Long Instruction Word CPU's get here, language implementers should be able to roll their own efficient type checking ops and I would expect the evolution toward Smalltalk to continue. -Mike * Sent from RemarQ http://www.*-*-*.com/ The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free!
|
Wed, 18 Jun 1902 08:00:00 GMT |
|
|
Frank A. Adria #5 / 8
|
smalltalk hasn't made it big
Quote: > When Long Instruction Word CPU's get here, language implementers > should be able to roll their own efficient type checking ops and I would > expect the evolution toward Smalltalk to continue.
Maybe. Maybe not. There IS a limit as to how much useful work you can get from a given C proc and it's relatively easy to shove simple type checking in those unused processor execution slots. But in the final analysis, it depends on how much loop unrolling and speculative execution possibilities can be got from the normal languages. As long as the static typers can get their extra 30% in performance, they will prevail in the marketplace (we all need faster screen savers, after all). More importantly, remember that the hardware guys will add only the amount of hardware to make SPECxxx run more efficiently. Any more than that (which might be used to help dynamically typed languages) would be viewed as wasted cost. And if you want a specialized memory subsystem (extra tag bits, etc.) - forget it! The performance gap will decrease, but as long as C and C++ remain the benchmark languages for hardware design, the dynamic languages will suffer. faa
|
Wed, 18 Jun 1902 08:00:00 GMT |
|
|
John Sarkel #6 / 8
|
smalltalk hasn't made it big
My experience teaching Smalltalk to a wide variety of people and consulting on a number of problematic projects suggests that there is a social/psychological barrier to the acceptance of Smalltalk. First, the patterns of collaborations that designers choose to use in systems under construction tend to reflect the social patterns of the development team. For a Smalltalk project to be deeply successful, the design must have objects that are communicative and that are designed to collaborate, not compete. Hmmmm. Further, to build reuseful frameworks, the objects that coordinate activities *MUST* delegate all of the actual work to pluggable helpers. Unfortunately, most of my corporate experience suggests that delegation and the release of control are novel practices in most offices. Second, Smalltalk is spare. I have had clients turn red faced and scream at me that they though OO was supposed to be easy. The source of their frustration was that they had truly difficult problems to solve. The problem with Smalltalk was that there was no elaborate syntax in which to get lost. Rather, as our analysis and design proceeded, the magnitude and difficulty of their very real problems started to emerge. My experience suggests that people find it psychologically easier to get lost in the denial of analysis paralysis than to face very real and difficult problems and attempt to find solutions. Oh well, that's this armchair philosphers quatre sous. John Sarkela "Speaking on my own and not representing any other interests." Quote:
> What has kept Smalltalk from catching on with the broader programming > community? Would you consider Smalltalk as the best OO language? I > recently read a book by Scott Ambler called 'Building Object Applications > That Work' and he convinced me (through words not experience) that > Smalltalk was the most pure of all OO programming languages. Are there any > drawbacks to the language as a result of the language being designed > relatively early to other programs? > What say you, present company? > Z
|
Wed, 18 Jun 1902 08:00:00 GMT |
|
|
pharsei #7 / 8
|
smalltalk hasn't made it big
I don't expect Smalltalk to execute as quickly as C, but I expect the gap to drop. The point about LIW CPU's is that they really don't target a particular language so much, but provide a bunch of simple ops. After reading The Design and Evaluation of a High Performance Smalltalk System it surpised me how little had to be done to get a big boost in Smalltalk runtime. And I expect that as processor speeds continue to increase, processing speed will matter less and programmer productivity will matter more. -Mike * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free!
|
Wed, 18 Jun 1902 08:00:00 GMT |
|
|
Reinout Heec #8 / 8
|
smalltalk hasn't made it big
Quote:
> After reading The Design and Evaluation of a High Performance Smalltalk System
I have some interest (personal education) in this subject, but see that this book is out of print http://mitpress.mit.edu/book-home.tcl?isbn=026221010X . Does anybody know any urls that cover the same subject/solution? TIA! Reinout Heeck -------------
|
Wed, 18 Jun 1902 08:00:00 GMT |
|
|
|